lfs.h revision 1.89
1 /* $NetBSD: lfs.h,v 1.89 2005/07/31 20:18:32 christos Exp $ */ 2 3 /*- 4 * Copyright (c) 1999, 2000, 2001, 2002, 2003 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Konrad E. Schroder <perseant (at) hhhh.org>. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. All advertising materials mentioning features or use of this software 19 * must display the following acknowledgement: 20 * This product includes software developed by the NetBSD 21 * Foundation, Inc. and its contributors. 22 * 4. Neither the name of The NetBSD Foundation nor the names of its 23 * contributors may be used to endorse or promote products derived 24 * from this software without specific prior written permission. 25 * 26 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 27 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 28 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 29 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 30 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 31 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 32 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 33 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 34 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 35 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 36 * POSSIBILITY OF SUCH DAMAGE. 37 */ 38 /*- 39 * Copyright (c) 1991, 1993 40 * The Regents of the University of California. All rights reserved. 41 * 42 * Redistribution and use in source and binary forms, with or without 43 * modification, are permitted provided that the following conditions 44 * are met: 45 * 1. Redistributions of source code must retain the above copyright 46 * notice, this list of conditions and the following disclaimer. 47 * 2. Redistributions in binary form must reproduce the above copyright 48 * notice, this list of conditions and the following disclaimer in the 49 * documentation and/or other materials provided with the distribution. 50 * 3. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 * 66 * @(#)lfs.h 8.9 (Berkeley) 5/8/95 67 */ 68 69 #ifndef _UFS_LFS_LFS_H_ 70 #define _UFS_LFS_LFS_H_ 71 72 /* 73 * Compile-time options for LFS. 74 */ 75 #define LFS_IFIND_RETRIES 16 76 #define LFS_LOGLENGTH 1024 /* size of debugging log */ 77 #define LFS_MAX_ACTIVE 10 /* Dirty segments before ckp forced */ 78 79 /* 80 * Fixed filesystem layout parameters 81 */ 82 #define LFS_LABELPAD 8192 /* LFS label size */ 83 #define LFS_SBPAD 8192 /* LFS superblock size */ 84 85 #define LFS_UNUSED_INUM 0 /* 0: out of band inode number */ 86 #define LFS_IFILE_INUM 1 /* 1: IFILE inode number */ 87 /* 2: Root inode number */ 88 #define LOSTFOUNDINO 3 /* 3: lost+found inode number */ 89 #define LFS_FIRST_INUM 4 /* 4: first free inode number */ 90 91 #define LFS_V1_SUMMARY_SIZE 512 /* V1 fixed summary size */ 92 #define LFS_DFL_SUMMARY_SIZE 512 /* Default summary size */ 93 94 #define LFS_MAX_DADDR 0x7fffffff /* Highest addressable fsb */ 95 96 /* Adjustable filesystem parameters */ 97 #define MIN_FREE_SEGS 2 98 #ifndef LFS_ATIME_IFILE 99 # define LFS_ATIME_IFILE 0 /* Store atime info in ifile (optional in LFSv1) */ 100 #endif 101 #define LFS_MARKV_MAXBLKCNT 65536 /* Max block count for lfs_markv() */ 102 103 /* Misc. definitions */ 104 #define BW_CLEAN 1 /* Flag for lfs_bwrite_ext() */ 105 #define PG_DELWRI PG_PAGER1 /* Local def for delayed pageout */ 106 107 /* Resource limits */ 108 #define LFS_MAX_BUFS ((nbuf >> 2) - 10) 109 #define LFS_WAIT_BUFS ((nbuf >> 1) - (nbuf >> 3) - 10) 110 #define LFS_INVERSE_MAX_BUFS(n) (((n) + 10) << 2) 111 #define LFS_MAX_BYTES ((bufmem_lowater >> 2) - 10 * PAGE_SIZE) 112 #define LFS_INVERSE_MAX_BYTES(n) (((n) + 10 * PAGE_SIZE) << 2) 113 #define LFS_WAIT_BYTES ((bufmem_lowater >> 1) - (bufmem_lowater >> 3) - 10 * PAGE_SIZE) 114 #define LFS_MAX_DIROP ((desiredvnodes >> 2) + (desiredvnodes >> 3)) 115 #define LFS_MAX_PAGES \ 116 (((uvmexp.active + uvmexp.inactive + uvmexp.free) * uvmexp.filemin) >> 8) 117 #define LFS_WAIT_PAGES \ 118 (((uvmexp.active + uvmexp.inactive + uvmexp.free) * uvmexp.filemax) >> 8) 119 #define LFS_BUFWAIT 2 /* How long to wait if over *_WAIT_* */ 120 121 /* 122 * Reserved blocks for lfs_malloc 123 */ 124 125 /* Structure to keep reserved blocks */ 126 typedef struct lfs_res_blk { 127 void *p; 128 LIST_ENTRY(lfs_res_blk) res; 129 int size; 130 char inuse; 131 } res_t; 132 133 /* Types for lfs_newbuf and lfs_malloc */ 134 #define LFS_NB_UNKNOWN -1 135 #define LFS_NB_SUMMARY 0 136 #define LFS_NB_SBLOCK 1 137 #define LFS_NB_IBLOCK 2 138 #define LFS_NB_CLUSTER 3 139 #define LFS_NB_CLEAN 4 140 #define LFS_NB_BLKIOV 5 141 #define LFS_NB_COUNT 6 /* always last */ 142 143 /* Number of reserved memory blocks of each type */ 144 #define LFS_N_SUMMARIES 2 145 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */ 146 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */ 147 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */ 148 #define LFS_N_CLEAN 0 149 #define LFS_N_BLKIOV 1 150 151 /* Total count of "large" (non-pool) types */ 152 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \ 153 LFS_N_CLUSTERS + LFS_N_CLEAN + LFS_N_BLKIOV) 154 155 /* Counts for pool types */ 156 #define LFS_N_CL LFS_N_CLUSTERS 157 #define LFS_N_BPP 2 158 #define LFS_N_SEG 2 159 160 /* 161 * "struct buf" associated definitions 162 */ 163 164 /* Unassigned disk addresses. */ 165 #define UNASSIGNED -1 166 #define UNWRITTEN -2 167 168 /* Unused logical block number */ 169 #define LFS_UNUSED_LBN -1 170 171 /* Determine if a buffer belongs to the ifile */ 172 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM) 173 174 # define LFS_LOCK_BUF(bp) do { \ 175 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \ 176 simple_lock(&lfs_subsys_lock); \ 177 ++locked_queue_count; \ 178 locked_queue_bytes += bp->b_bufsize; \ 179 simple_unlock(&lfs_subsys_lock); \ 180 } \ 181 (bp)->b_flags |= B_LOCKED; \ 182 } while (0) 183 184 # define LFS_UNLOCK_BUF(bp) do { \ 185 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \ 186 simple_lock(&lfs_subsys_lock); \ 187 --locked_queue_count; \ 188 locked_queue_bytes -= bp->b_bufsize; \ 189 if (locked_queue_count < LFS_WAIT_BUFS && \ 190 locked_queue_bytes < LFS_WAIT_BYTES) \ 191 wakeup(&locked_queue_count); \ 192 simple_unlock(&lfs_subsys_lock); \ 193 } \ 194 (bp)->b_flags &= ~B_LOCKED; \ 195 } while (0) 196 197 #ifdef _KERNEL 198 199 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */ 200 201 # define LFS_IS_MALLOC_BUF(bp) (((bp)->b_flags & B_CALL) && \ 202 (bp)->b_iodone == lfs_callback) 203 204 # ifdef DEBUG 205 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 206 if (lfs_debug_log_subsys[DLOG_LLIST]) { \ 207 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 208 wakeup(&locked_queue_count); \ 209 } \ 210 } while (0) 211 # else 212 # define LFS_DEBUG_COUNTLOCKED(m) 213 # endif 214 215 /* log for debugging writes to the Ifile */ 216 # ifdef DEBUG 217 struct lfs_log_entry { 218 const char *op; 219 const char *file; 220 int pid; 221 int line; 222 daddr_t block; 223 unsigned long flags; 224 }; 225 extern int lfs_lognum; 226 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 227 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 228 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\ 229 int _s; \ 230 \ 231 simple_lock(&lfs_subsys_lock); \ 232 _s = splbio(); \ 233 lfs_log[lfs_lognum].op = theop; \ 234 lfs_log[lfs_lognum].file = thefile; \ 235 lfs_log[lfs_lognum].line = (theline); \ 236 lfs_log[lfs_lognum].pid = (thepid); \ 237 lfs_log[lfs_lognum].block = (lbn); \ 238 lfs_log[lfs_lognum].flags = (theflags); \ 239 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 240 splx(_s); \ 241 simple_unlock(&lfs_subsys_lock); \ 242 } while (0) 243 244 # define LFS_BCLEAN_LOG(fs, bp) do { \ 245 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 246 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \ 247 bp->b_lblkno, bp->b_flags, curproc->p_pid);\ 248 } while (0) 249 250 /* Must match list in lfs_vfsops.c ! */ 251 # define DLOG_RF 0 /* roll forward */ 252 # define DLOG_ALLOC 1 /* inode alloc */ 253 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */ 254 # define DLOG_FLUSH 3 /* flush */ 255 # define DLOG_LLIST 4 /* locked list accounting */ 256 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */ 257 # define DLOG_VNODE 6 /* vflush/writevnodes */ 258 # define DLOG_SEG 7 /* segwrite */ 259 # define DLOG_SU 8 /* seguse accounting */ 260 # define DLOG_CLEAN 9 /* cleaner routines */ 261 # define DLOG_MOUNT 10 /* mount/unmount */ 262 # define DLOG_PAGE 11 /* putpages/gop_write */ 263 # define DLOG_DIROP 12 /* dirop accounting */ 264 # define DLOG_MALLOC 13 /* lfs_malloc accounting */ 265 # define DLOG_MAX 14 /* The terminator */ 266 # define DLOG(a) lfs_debug_log a 267 # else /* ! DEBUG */ 268 # define LFS_BCLEAN_LOG(fs, bp) 269 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 270 # define DLOG(a) 271 # endif /* ! DEBUG */ 272 #else /* ! _KERNEL */ 273 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 274 #endif /* _KERNEL */ 275 276 #ifdef _KERNEL 277 /* Filehandle structure for exported LFSes */ 278 struct lfid { 279 struct ufid lfid_ufid; 280 #define lfid_len lfid_ufid.ufid_len 281 #define lfid_ino lfid_ufid.ufid_ino 282 #define lfid_gen lfid_ufid.ufid_gen 283 uint32_t lfid_ident; 284 }; 285 #endif /* _KERNEL */ 286 287 /* 288 * "struct inode" associated definitions 289 */ 290 291 /* Address calculations for metadata located in the inode */ 292 #define S_INDIR(fs) -NDADDR 293 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 294 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 295 296 /* For convenience */ 297 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING) 298 299 #define LFS_SET_UINO(ip, flags) do { \ 300 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 301 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \ 302 ++(ip)->i_lfs->lfs_uinodes; \ 303 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \ 304 ++(ip)->i_lfs->lfs_uinodes; \ 305 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \ 306 ++(ip)->i_lfs->lfs_uinodes; \ 307 (ip)->i_flag |= (flags); \ 308 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 309 } while (0) 310 311 #define LFS_CLR_UINO(ip, flags) do { \ 312 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 313 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 314 --(ip)->i_lfs->lfs_uinodes; \ 315 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 316 --(ip)->i_lfs->lfs_uinodes; \ 317 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 318 --(ip)->i_lfs->lfs_uinodes; \ 319 (ip)->i_flag &= ~(flags); \ 320 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 321 panic("lfs_uinodes < 0"); \ 322 } \ 323 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 324 } while (0) 325 326 #define LFS_ITIMES(ip, acc, mod, cre) do { \ 327 struct lfs *_fs = (ip)->i_lfs; \ 328 \ 329 if ((ip)->i_flag & IN_ACCESS) { \ 330 (ip)->i_ffs1_atime = (acc)->tv_sec; \ 331 (ip)->i_ffs1_atimensec = (acc)->tv_nsec; \ 332 if ((ip)->i_lfs->lfs_version > 1) { \ 333 struct buf *_ibp; \ 334 IFILE *_ifp; \ 335 \ 336 LFS_IENTRY(_ifp, ip->i_lfs, ip->i_number, _ibp); \ 337 _ifp->if_atime_sec = (acc)->tv_sec; \ 338 _ifp->if_atime_nsec = (acc)->tv_nsec; \ 339 LFS_BWRITE_LOG(_ibp); \ 340 simple_lock(&_fs->lfs_interlock); \ 341 _fs->lfs_flags |= LFS_IFDIRTY; \ 342 simple_unlock(&_fs->lfs_interlock); \ 343 } else { \ 344 LFS_SET_UINO(ip, IN_ACCESSED); \ 345 } \ 346 } \ 347 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFY)) { \ 348 if ((ip)->i_flag & (IN_UPDATE | IN_MODIFY)) { \ 349 (ip)->i_ffs1_mtime = (mod)->tv_sec; \ 350 (ip)->i_ffs1_mtimensec = (mod)->tv_nsec; \ 351 (ip)->i_modrev++; \ 352 } \ 353 if ((ip)->i_flag & (IN_CHANGE | IN_MODIFY)) { \ 354 (ip)->i_ffs1_ctime = (cre)->tv_sec; \ 355 (ip)->i_ffs1_ctimensec = (cre)->tv_nsec; \ 356 } \ 357 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE)) \ 358 LFS_SET_UINO(ip, IN_MODIFIED); \ 359 if ((ip)->i_flag & IN_MODIFY) \ 360 LFS_SET_UINO(ip, IN_ACCESSED); \ 361 } \ 362 (ip)->i_flag &= ~(IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY);\ 363 } while (0) 364 365 /* 366 * "struct vnode" associated definitions 367 */ 368 369 /* Heuristic emptiness measure */ 370 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 371 !((vp)->v_flag & VONWORKLST) && \ 372 VTOI(vp)->i_lfs_nbtree == 0) 373 374 /* XXX Shouldn't we use v_numoutput instead? */ 375 #define WRITEINPROG(vp) (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 376 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING))) 377 378 379 /* 380 * On-disk and in-memory checkpoint segment usage structure. 381 */ 382 typedef struct segusage SEGUSE; 383 struct segusage { 384 u_int32_t su_nbytes; /* 0: number of live bytes */ 385 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 386 u_int16_t su_nsums; /* 8: number of summaries in segment */ 387 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 388 389 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 390 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 391 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 392 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 393 #define SEGUSE_EMPTY 0x10 /* segment is empty */ 394 #define SEGUSE_INVAL 0x20 /* segment is invalid */ 395 u_int32_t su_flags; /* 12: segment flags */ 396 u_int64_t su_lastmod; /* 16: last modified timestamp */ 397 }; 398 399 typedef struct segusage_v1 SEGUSE_V1; 400 struct segusage_v1 { 401 u_int32_t su_nbytes; /* 0: number of live bytes */ 402 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 403 u_int16_t su_nsums; /* 8: number of summaries in segment */ 404 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 405 u_int32_t su_flags; /* 12: segment flags */ 406 }; 407 408 #define SEGUPB(fs) (fs->lfs_sepb) 409 #define SEGTABSIZE_SU(fs) \ 410 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb) 411 412 #ifdef _KERNEL 413 # define SHARE_IFLOCK(F) \ 414 do { \ 415 simple_lock(&(F)->lfs_interlock); \ 416 lockmgr(&(F)->lfs_iflock, LK_SHARED, &(F)->lfs_interlock); \ 417 simple_unlock(&(F)->lfs_interlock); \ 418 } while(0) 419 # define UNSHARE_IFLOCK(F) \ 420 do { \ 421 simple_lock(&(F)->lfs_interlock); \ 422 lockmgr(&(F)->lfs_iflock, LK_RELEASE, &(F)->lfs_interlock); \ 423 simple_unlock(&(F)->lfs_interlock); \ 424 } while(0) 425 #else /* ! _KERNEL */ 426 # define SHARE_IFLOCK(F) 427 # define UNSHARE_IFLOCK(F) 428 #endif /* ! _KERNEL */ 429 430 /* Read in the block with a specific segment usage entry from the ifile. */ 431 #define LFS_SEGENTRY(SP, F, IN, BP) do { \ 432 int _e; \ 433 SHARE_IFLOCK(F); \ 434 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 435 if ((_e = bread((F)->lfs_ivnode, \ 436 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 437 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 438 panic("lfs: ifile read: %d", _e); \ 439 if ((F)->lfs_version == 1) \ 440 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 441 ((IN) & ((F)->lfs_sepb - 1))); \ 442 else \ 443 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 444 UNSHARE_IFLOCK(F); \ 445 } while (0) 446 447 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \ 448 if ((SP)->su_nbytes == 0) \ 449 (SP)->su_flags |= SEGUSE_EMPTY; \ 450 else \ 451 (SP)->su_flags &= ~SEGUSE_EMPTY; \ 452 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \ 453 LFS_BWRITE_LOG(BP); \ 454 } while (0) 455 456 /* 457 * On-disk file information. One per file with data blocks in the segment. 458 */ 459 typedef struct finfo FINFO; 460 struct finfo { 461 u_int32_t fi_nblocks; /* number of blocks */ 462 u_int32_t fi_version; /* version number */ 463 u_int32_t fi_ino; /* inode number */ 464 u_int32_t fi_lastlength; /* length of last block in array */ 465 int32_t fi_blocks[1]; /* array of logical block numbers */ 466 }; 467 /* sizeof FINFO except fi_blocks */ 468 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t)) 469 470 /* 471 * Index file inode entries. 472 */ 473 typedef struct ifile IFILE; 474 struct ifile { 475 u_int32_t if_version; /* inode version number */ 476 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 477 int32_t if_daddr; /* inode disk address */ 478 ino_t if_nextfree; /* next-unallocated inode */ 479 u_int32_t if_atime_sec; /* Last access time, seconds */ 480 u_int32_t if_atime_nsec; /* and nanoseconds */ 481 }; 482 483 typedef struct ifile_v1 IFILE_V1; 484 struct ifile_v1 { 485 u_int32_t if_version; /* inode version number */ 486 int32_t if_daddr; /* inode disk address */ 487 ino_t if_nextfree; /* next-unallocated inode */ 488 #if LFS_ATIME_IFILE 489 struct timespec if_atime; /* Last access time */ 490 #endif 491 }; 492 493 /* 494 * LFSv1 compatibility code is not allowed to touch if_atime, since it 495 * may not be mapped! 496 */ 497 /* Read in the block with a specific inode from the ifile. */ 498 #define LFS_IENTRY(IP, F, IN, BP) do { \ 499 int _e; \ 500 SHARE_IFLOCK(F); \ 501 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 502 if ((_e = bread((F)->lfs_ivnode, \ 503 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 504 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 505 panic("lfs: ifile read %d", _e); \ 506 if ((F)->lfs_version == 1) \ 507 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \ 508 (IN) % (F)->lfs_ifpb); \ 509 else \ 510 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 511 UNSHARE_IFLOCK(F); \ 512 } while (0) 513 514 /* 515 * Cleaner information structure. This resides in the ifile and is used 516 * to pass information from the kernel to the cleaner. 517 */ 518 typedef struct _cleanerinfo { 519 u_int32_t clean; /* number of clean segments */ 520 u_int32_t dirty; /* number of dirty segments */ 521 int32_t bfree; /* disk blocks free */ 522 int32_t avail; /* disk blocks available */ 523 u_int32_t free_head; /* head of the inode free list */ 524 u_int32_t free_tail; /* tail of the inode free list */ 525 } CLEANERINFO; 526 527 #define CLEANSIZE_SU(fs) \ 528 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift) 529 530 /* Read in the block with the cleaner info from the ifile. */ 531 #define LFS_CLEANERINFO(CP, F, BP) do { \ 532 SHARE_IFLOCK(F); \ 533 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 534 if (bread((F)->lfs_ivnode, \ 535 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \ 536 panic("lfs: ifile read"); \ 537 (CP) = (CLEANERINFO *)(BP)->b_data; \ 538 UNSHARE_IFLOCK(F); \ 539 } while (0) 540 541 /* 542 * Synchronize the Ifile cleaner info with current avail and bfree. 543 */ 544 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \ 545 simple_lock(&(fs)->lfs_interlock); \ 546 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \ 547 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \ 548 (fs)->lfs_favail) { \ 549 (cip)->bfree = (fs)->lfs_bfree; \ 550 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \ 551 (fs)->lfs_favail; \ 552 if (((bp)->b_flags & B_GATHERED) == 0) { \ 553 (fs)->lfs_flags |= LFS_IFDIRTY; \ 554 } \ 555 simple_unlock(&(fs)->lfs_interlock); \ 556 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \ 557 } else { \ 558 simple_unlock(&(fs)->lfs_interlock); \ 559 brelse(bp); \ 560 } \ 561 } while (0) 562 563 /* 564 * Get the head of the inode free list. 565 * Always caled with the segment lock held. 566 */ 567 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \ 568 if ((FS)->lfs_version > 1) { \ 569 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 570 (FS)->lfs_freehd = (CIP)->free_head; \ 571 brelse(BP); \ 572 } \ 573 *(FREEP) = (FS)->lfs_freehd; \ 574 } while (0) 575 576 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \ 577 (FS)->lfs_freehd = (VAL); \ 578 if ((FS)->lfs_version > 1) { \ 579 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 580 (CIP)->free_head = (VAL); \ 581 LFS_BWRITE_LOG(BP); \ 582 simple_lock(&fs->lfs_interlock); \ 583 (FS)->lfs_flags |= LFS_IFDIRTY; \ 584 simple_unlock(&fs->lfs_interlock); \ 585 } \ 586 } while (0) 587 588 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \ 589 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 590 *(FREEP) = (CIP)->free_tail; \ 591 brelse(BP); \ 592 } while (0) 593 594 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \ 595 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 596 (CIP)->free_tail = (VAL); \ 597 LFS_BWRITE_LOG(BP); \ 598 simple_lock(&fs->lfs_interlock); \ 599 (FS)->lfs_flags |= LFS_IFDIRTY; \ 600 simple_unlock(&fs->lfs_interlock); \ 601 } while (0) 602 603 /* 604 * On-disk segment summary information 605 */ 606 typedef struct segsum_v1 SEGSUM_V1; 607 struct segsum_v1 { 608 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 609 u_int32_t ss_datasum; /* 4: check sum of data */ 610 u_int32_t ss_magic; /* 8: segment summary magic number */ 611 #define SS_MAGIC 0x061561 612 int32_t ss_next; /* 12: next segment */ 613 u_int32_t ss_create; /* 16: creation time stamp */ 614 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 615 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 616 617 #define SS_DIROP 0x01 /* segment begins a dirop */ 618 #define SS_CONT 0x02 /* more partials to finish this write*/ 619 u_int16_t ss_flags; /* 24: used for directory operations */ 620 u_int16_t ss_pad; /* 26: extra space */ 621 /* FINFO's and inode daddr's... */ 622 }; 623 624 typedef struct segsum SEGSUM; 625 struct segsum { 626 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 627 u_int32_t ss_datasum; /* 4: check sum of data */ 628 u_int32_t ss_magic; /* 8: segment summary magic number */ 629 int32_t ss_next; /* 12: next segment */ 630 u_int32_t ss_ident; /* 16: roll-forward fsid */ 631 #define ss_ocreate ss_ident /* ident is where create was in v1 */ 632 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 633 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 634 u_int16_t ss_flags; /* 24: used for directory operations */ 635 u_int8_t ss_pad[6]; /* 26: extra space */ 636 u_int64_t ss_serial; /* 32: serial number */ 637 u_int64_t ss_create; /* 40: time stamp */ 638 /* FINFO's and inode daddr's... */ 639 }; 640 641 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM)) 642 643 644 /* 645 * On-disk super block. 646 */ 647 struct dlfs { 648 #define LFS_MAGIC 0x070162 649 u_int32_t dlfs_magic; /* 0: magic number */ 650 #define LFS_VERSION 2 651 u_int32_t dlfs_version; /* 4: version number */ 652 653 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 654 /* number of frags in fs (v2) */ 655 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 656 /* number of bytes per segment (v2) */ 657 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 658 u_int32_t dlfs_bsize; /* 20: file system block size */ 659 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 660 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */ 661 662 /* Checkpoint region. */ 663 u_int32_t dlfs_freehd; /* 32: start of the free list */ 664 int32_t dlfs_bfree; /* 36: number of free disk blocks */ 665 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */ 666 int32_t dlfs_avail; /* 44: blocks available for writing */ 667 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 668 int32_t dlfs_idaddr; /* 52: inode file disk address */ 669 u_int32_t dlfs_ifile; /* 56: inode file inode number */ 670 int32_t dlfs_lastseg; /* 60: address of last segment written */ 671 int32_t dlfs_nextseg; /* 64: address of next segment to write */ 672 int32_t dlfs_curseg; /* 68: current segment being written */ 673 int32_t dlfs_offset; /* 72: offset in curseg for next partial */ 674 int32_t dlfs_lastpseg; /* 76: address of last partial written */ 675 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 676 #define dlfs_otstamp dlfs_inopf 677 678 /* These are configuration parameters. */ 679 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 680 681 /* These fields can be computed from the others. */ 682 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 683 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */ 684 u_int32_t dlfs_inopb; /* 100: inodes per block */ 685 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */ 686 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 687 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */ 688 u_int32_t dlfs_nseg; /* 116: number of segments */ 689 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */ 690 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 691 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 692 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */ 693 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */ 694 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */ 695 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 696 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 697 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */ 698 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 699 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 700 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 701 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 702 703 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 704 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */ 705 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */ 706 int32_t dlfs_sboffs[LFS_MAXNUMSB]; 707 708 u_int32_t dlfs_nclean; /* 228: Number of clean segments */ 709 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 710 #define LFS_PF_CLEAN 0x1 711 u_int16_t dlfs_pflags; /* 322: file system persistent flags */ 712 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 713 u_int32_t dlfs_minfreeseg; /* 328: segs reserved for cleaner */ 714 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */ 715 u_int64_t dlfs_serial; /* 336: serial number */ 716 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */ 717 int32_t dlfs_start; /* 348: start of segment 0 */ 718 u_int64_t dlfs_tstamp; /* 352: time stamp */ 719 #define LFS_44INODEFMT 0 720 #define LFS_MAXINODEFMT 0 721 u_int32_t dlfs_inodefmt; /* 360: inode format version */ 722 u_int32_t dlfs_interleave; /* 364: segment interleave */ 723 u_int32_t dlfs_ident; /* 368: per-fs identifier */ 724 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */ 725 int8_t dlfs_pad[132]; /* 376: round to 512 bytes */ 726 /* Checksum -- last valid disk field. */ 727 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */ 728 }; 729 730 /* 731 * In-memory super block. 732 */ 733 struct lfs { 734 struct dlfs lfs_dlfs; /* on-disk parameters */ 735 #define lfs_magic lfs_dlfs.dlfs_magic 736 #define lfs_version lfs_dlfs.dlfs_version 737 #define lfs_size lfs_dlfs.dlfs_size 738 #define lfs_ssize lfs_dlfs.dlfs_ssize 739 #define lfs_dsize lfs_dlfs.dlfs_dsize 740 #define lfs_bsize lfs_dlfs.dlfs_bsize 741 #define lfs_fsize lfs_dlfs.dlfs_fsize 742 #define lfs_frag lfs_dlfs.dlfs_frag 743 #define lfs_freehd lfs_dlfs.dlfs_freehd 744 #define lfs_bfree lfs_dlfs.dlfs_bfree 745 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 746 #define lfs_avail lfs_dlfs.dlfs_avail 747 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 748 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 749 #define lfs_ifile lfs_dlfs.dlfs_ifile 750 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 751 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 752 #define lfs_curseg lfs_dlfs.dlfs_curseg 753 #define lfs_offset lfs_dlfs.dlfs_offset 754 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 755 #define lfs_otstamp lfs_dlfs.dlfs_inopf 756 #define lfs_inopf lfs_dlfs.dlfs_inopf 757 #define lfs_minfree lfs_dlfs.dlfs_minfree 758 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 759 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 760 #define lfs_inopb lfs_dlfs.dlfs_inopb 761 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 762 #define lfs_sepb lfs_dlfs.dlfs_sepb 763 #define lfs_nindir lfs_dlfs.dlfs_nindir 764 #define lfs_nseg lfs_dlfs.dlfs_nseg 765 #define lfs_nspf lfs_dlfs.dlfs_nspf 766 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 767 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 768 #define lfs_segmask lfs_dlfs.dlfs_segmask 769 #define lfs_segshift lfs_dlfs.dlfs_segshift 770 #define lfs_bmask lfs_dlfs.dlfs_bmask 771 #define lfs_bshift lfs_dlfs.dlfs_bshift 772 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 773 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 774 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 775 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 776 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 777 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 778 #define lfs_sushift lfs_dlfs.dlfs_sushift 779 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 780 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 781 #define lfs_cksum lfs_dlfs.dlfs_cksum 782 #define lfs_pflags lfs_dlfs.dlfs_pflags 783 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 784 #define lfs_nclean lfs_dlfs.dlfs_nclean 785 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 786 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 787 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 788 #define lfs_serial lfs_dlfs.dlfs_serial 789 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 790 #define lfs_start lfs_dlfs.dlfs_start 791 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 792 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 793 #define lfs_interleave lfs_dlfs.dlfs_interleave 794 #define lfs_ident lfs_dlfs.dlfs_ident 795 796 /* These fields are set at mount time and are meaningless on disk. */ 797 struct segment *lfs_sp; /* current segment being written */ 798 struct vnode *lfs_ivnode; /* vnode for the ifile */ 799 u_int32_t lfs_seglock; /* single-thread the segment writer */ 800 pid_t lfs_lockpid; /* pid of lock holder */ 801 u_int32_t lfs_iocount; /* number of ios pending */ 802 u_int32_t lfs_writer; /* don't allow any dirops to start */ 803 u_int32_t lfs_dirops; /* count of active directory ops */ 804 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 805 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 806 int8_t lfs_fmod; /* super block modified flag */ 807 int8_t lfs_ronly; /* mounted read-only flag */ 808 #define LFS_NOTYET 0x01 809 #define LFS_IFDIRTY 0x02 810 #define LFS_WARNED 0x04 811 #define LFS_UNDIROP 0x08 812 int8_t lfs_flags; /* currently unused flag */ 813 u_int16_t lfs_activesb; /* toggle between superblocks */ 814 daddr_t lfs_sbactive; /* disk address of current sb write */ 815 struct vnode *lfs_flushvp; /* vnode being flushed */ 816 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 817 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 818 size_t lfs_devbsize; /* Device block size */ 819 size_t lfs_devbshift; /* Device block shift */ 820 struct lock lfs_fraglock; 821 struct lock lfs_iflock; /* Ifile lock */ 822 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 823 int lfs_nadirop; /* number of active dirop nodes */ 824 long lfs_ravail; /* blocks pre-reserved for writing */ 825 long lfs_favail; /* blocks pre-reserved for writing */ 826 res_t *lfs_resblk; /* Reserved memory for pageout */ 827 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */ 828 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */ 829 #define LFS_RESHASH_WIDTH 17 830 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH]; 831 int lfs_pdflush; /* pagedaemon wants us to flush */ 832 u_int32_t **lfs_suflags; /* Segment use flags */ 833 #ifdef _KERNEL 834 struct pool lfs_clpool; /* Pool for struct lfs_cluster */ 835 struct pool lfs_bpppool; /* Pool for bpp */ 836 struct pool lfs_segpool; /* Pool for struct segment */ 837 #endif /* _KERNEL */ 838 #define LFS_MAX_CLEANIND 64 839 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */ 840 int lfs_cleanind; /* Index into intervals */ 841 struct simplelock lfs_interlock; /* lock for lfs_seglock */ 842 int lfs_sleepers; /* # procs sleeping this fs */ 843 int lfs_pages; /* dirty pages blaming this fs */ 844 }; 845 846 /* NINDIR is the number of indirects in a file system block. */ 847 #define NINDIR(fs) ((fs)->lfs_nindir) 848 849 /* INOPB is the number of inodes in a secondary storage block. */ 850 #define INOPB(fs) ((fs)->lfs_inopb) 851 /* INOPF is the number of inodes in a fragment. */ 852 #define INOPF(fs) ((fs)->lfs_inopf) 853 854 #define blksize(fs, ip, lbn) \ 855 (((lbn) >= NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \ 856 ? (fs)->lfs_bsize \ 857 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size)))) 858 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask)) 859 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \ 860 ((int)((loc) & (fs)->lfs_ffmask)) 861 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb) 862 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb) 863 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 864 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 865 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift) 866 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift) 867 /* Same as above, but named like dbtob(), btodb() */ 868 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \ 869 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 870 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \ 871 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 872 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 873 (fs)->lfs_fsbtodb)) 874 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 875 (fs)->lfs_fsbtodb)) 876 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift) 877 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \ 878 ((loc) >> (fs)->lfs_ffshift) 879 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \ 880 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask))) 881 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \ 882 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask))) 883 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \ 884 ((frags) >> (fs)->lfs_fbshift) 885 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \ 886 ((blks) << (fs)->lfs_fbshift) 887 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \ 888 ((fsb) & ((fs)->lfs_frag - 1)) 889 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \ 890 ((fsb) &~ ((fs)->lfs_frag - 1)) 891 #define dblksize(fs, dp, lbn) \ 892 (((lbn) >= NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\ 893 ? (fs)->lfs_bsize \ 894 : (fragroundup(fs, blkoff(fs, (dp)->di_size)))) 895 896 #define segsize(fs) ((fs)->lfs_version == 1 ? \ 897 lblktosize((fs), (fs)->lfs_ssize) : \ 898 (fs)->lfs_ssize) 899 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \ 900 (fs)->lfs_ssize << (fs)->lfs_blktodb : \ 901 btofsb((fs), (fs)->lfs_ssize)) * (seg)) 902 #define dtosn(fs, daddr) /* block address to segment number */ \ 903 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1))) 904 #define sntod(fs, sn) /* segment number to disk address */ \ 905 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start)) 906 907 /* 908 * Structures used by lfs_bmapv and lfs_markv to communicate information 909 * about inodes and data blocks. 910 */ 911 typedef struct block_info { 912 ino_t bi_inode; /* inode # */ 913 int32_t bi_lbn; /* logical block w/in file */ 914 int32_t bi_daddr; /* disk address of block */ 915 u_int64_t bi_segcreate; /* origin segment create time */ 916 int bi_version; /* file version number */ 917 void *bi_bp; /* data buffer */ 918 int bi_size; /* size of the block (if fragment) */ 919 } BLOCK_INFO; 920 921 /* Compatibility for 1.5 binaries */ 922 typedef struct block_info_15 { 923 ino_t bi_inode; /* inode # */ 924 int32_t bi_lbn; /* logical block w/in file */ 925 int32_t bi_daddr; /* disk address of block */ 926 u_int32_t bi_segcreate; /* origin segment create time */ 927 int bi_version; /* file version number */ 928 void *bi_bp; /* data buffer */ 929 int bi_size; /* size of the block (if fragment) */ 930 } BLOCK_INFO_15; 931 932 /* In-memory description of a segment about to be written. */ 933 struct segment { 934 struct lfs *fs; /* file system pointer */ 935 struct buf **bpp; /* pointer to buffer array */ 936 struct buf **cbpp; /* pointer to next available bp */ 937 struct buf **start_bpp; /* pointer to first bp in this set */ 938 struct buf *ibp; /* buffer pointer to inode page */ 939 struct ufs1_dinode *idp; /* pointer to ifile dinode */ 940 struct finfo *fip; /* current fileinfo pointer */ 941 struct vnode *vp; /* vnode being gathered */ 942 void *segsum; /* segment summary info */ 943 u_int32_t ninodes; /* number of inodes in this segment */ 944 int32_t seg_bytes_left; /* bytes left in segment */ 945 int32_t sum_bytes_left; /* bytes left in summary block */ 946 u_int32_t seg_number; /* number of this segment */ 947 int32_t *start_lbp; /* beginning lbn for this set */ 948 949 #define SEGM_CKP 0x01 /* doing a checkpoint */ 950 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */ 951 #define SEGM_SYNC 0x04 /* wait for segment */ 952 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */ 953 #define SEGM_PAGEDAEMON 0x10 /* pagedaemon called us */ 954 #define SEGM_WRITERD 0x20 /* LFS writed called us */ 955 #define SEGM_FORCE_CKP 0x40 /* Force checkpoint right away */ 956 u_int16_t seg_flags; /* run-time flags for this segment */ 957 u_int32_t seg_iocount; /* number of ios pending */ 958 int ndupino; /* number of duplicate inodes */ 959 }; 960 961 #ifdef _KERNEL 962 struct lfs_cluster { 963 size_t bufsize; /* Size of kept data */ 964 struct buf **bpp; /* Array of kept buffers */ 965 int bufcount; /* Number of kept buffers */ 966 #define LFS_CL_MALLOC 0x00000001 967 #define LFS_CL_SHIFT 0x00000002 968 #define LFS_CL_SYNC 0x00000004 969 u_int32_t flags; /* Flags */ 970 struct lfs *fs; /* LFS that this belongs to */ 971 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */ 972 }; 973 974 /* 975 * Splay tree containing block numbers allocated through lfs_balloc. 976 */ 977 struct lbnentry { 978 SPLAY_ENTRY(lbnentry) entry; 979 daddr_t lbn; 980 }; 981 #endif /* _KERNEL */ 982 983 /* 984 * LFS inode extensions. 985 */ 986 struct lfs_inode_ext { 987 off_t lfs_osize; /* size of file on disk */ 988 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */ 989 size_t lfs_fragsize[NDADDR]; /* size of on-disk direct blocks */ 990 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */ 991 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */ 992 #define LFSI_NO_GOP_WRITE 0x01 993 u_int32_t lfs_iflags; /* Inode flags */ 994 daddr_t lfs_hiblk; /* Highest lbn held by inode */ 995 #ifdef _KERNEL 996 SPLAY_HEAD(lfs_splay, lbnentry) lfs_lbtree; /* Tree of balloc'd lbns */ 997 int lfs_nbtree; /* Size of tree */ 998 #endif 999 }; 1000 #define i_lfs_osize inode_ext.lfs->lfs_osize 1001 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks 1002 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize 1003 #define i_lfs_dchain inode_ext.lfs->lfs_dchain 1004 #define i_lfs_iflags inode_ext.lfs->lfs_iflags 1005 #define i_lfs_hiblk inode_ext.lfs->lfs_hiblk 1006 #define i_lfs_lbtree inode_ext.lfs->lfs_lbtree 1007 #define i_lfs_nbtree inode_ext.lfs->lfs_nbtree 1008 1009 /* 1010 * Macros for determining free space on the disk, with the variable metadata 1011 * of segment summaries and inode blocks taken into account. 1012 */ 1013 /* Estimate number of clean blocks not available for writing */ 1014 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \ 1015 (int64_t)(F)->lfs_nclean) / \ 1016 ((F)->lfs_nseg - (F)->lfs_nclean))) 1017 1018 /* Estimate total size of the disk not including metadata */ 1019 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 1020 1021 /* Estimate number of blocks actually available for writing */ 1022 #define LFS_EST_BFREE(F) ((F)->lfs_bfree > LFS_EST_CMETA(F) + (F)->lfs_dmeta ? (F)->lfs_bfree - LFS_EST_CMETA(F) - (F)->lfs_dmeta : 0) 1023 1024 /* Amount of non-meta space not available to mortal man */ 1025 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 1026 (u_int64_t)(F)->lfs_minfree) / \ 1027 100) 1028 1029 /* Can credential C write BB blocks */ 1030 #define ISSPACE(F, BB, C) \ 1031 ((((C) == NOCRED || (C)->cr_uid == 0) && \ 1032 LFS_EST_BFREE(F) >= (BB)) || \ 1033 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB))) 1034 1035 /* Can an ordinary user write BB blocks */ 1036 #define IS_FREESPACE(F, BB) \ 1037 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 1038 1039 /* 1040 * The minimum number of blocks to create a new inode. This is: 1041 * directory direct block (1) + NIADDR indirect blocks + inode block (1) + 1042 * ifile direct block (1) + NIADDR indirect blocks = 3 + 2 * NIADDR blocks. 1043 */ 1044 #define LFS_NRESERVE(F) (btofsb((F), (2 * NIADDR + 3) << (F)->lfs_bshift)) 1045 1046 /* Statistics Counters */ 1047 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1048 u_int segsused; 1049 u_int psegwrites; 1050 u_int psyncwrites; 1051 u_int pcleanwrites; 1052 u_int blocktot; 1053 u_int cleanblocks; 1054 u_int ncheckpoints; 1055 u_int nwrites; 1056 u_int nsync_writes; 1057 u_int wait_exceeded; 1058 u_int write_exceeded; 1059 u_int flush_invoked; 1060 u_int vflush_invoked; 1061 u_int clean_inlocked; 1062 u_int clean_vnlocked; 1063 u_int segs_reclaimed; 1064 }; 1065 #ifdef _KERNEL 1066 extern struct lfs_stats lfs_stats; 1067 #endif 1068 1069 /* Fcntls to take the place of the lfs syscalls */ 1070 struct lfs_fcntl_markv { 1071 BLOCK_INFO *blkiov; /* blocks to relocate */ 1072 int blkcnt; /* number of blocks */ 1073 }; 1074 1075 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 0, struct timeval) 1076 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 1, struct timeval) 1077 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv) 1078 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv) 1079 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4) 1080 #define LFCNIFILEFH _FCNW_FSPRIV('L', 5, struct fhandle) 1081 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int) 1082 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int) 1083 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int) 1084 /* Compat for NetBSD 2.x bug */ 1085 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval) 1086 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval) 1087 1088 #ifdef _KERNEL 1089 /* XXX MP */ 1090 #define LFS_SEGLOCK_HELD(fs) \ 1091 ((fs)->lfs_seglock != 0 && (fs)->lfs_lockpid == curproc->p_pid) 1092 #endif /* _KERNEL */ 1093 1094 /* Debug segment lock */ 1095 #ifdef notyet 1096 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1097 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1098 # define ASSERT_DUNNO_SEGLOCK(fs) 1099 # define ASSERT_MAYBE_SEGLOCK(fs) 1100 #else /* !notyet */ 1101 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1102 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1103 LFS_SEGLOCK_HELD(fs))) 1104 # define ASSERT_SEGLOCK(fs) do { \ 1105 if (!LFS_SEGLOCK_HELD(fs)) { \ 1106 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1107 } \ 1108 } while(0) 1109 # define ASSERT_NO_SEGLOCK(fs) do { \ 1110 if (LFS_SEGLOCK_HELD(fs)) { \ 1111 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1112 } \ 1113 } while(0) 1114 # define ASSERT_MAYBE_SEGLOCK(x) 1115 #endif /* !notyet */ 1116 1117 #endif /* !_UFS_LFS_LFS_H_ */ 1118
Indexes created Tue Sep 30 17:09:57 GMT 2025